The OpenStack Compute service allows you to control an Infrastructure-as-a-Service (IaaS) cloud computing platform. It gives you control over instances and networks, and allows you to manage access to the cloud through users and projects.
Compute does not include virtualization software. Instead, it defines drivers that interact with underlying virtualization mechanisms that run on your host operating system, and exposes functionality over a web-based API.
To effectively administer compute, you must understand how the different installed nodes interact with each other. Compute can be installed in many different ways using multiple servers, but generally multiple compute nodes control the virtual servers and a cloud controller node contains the remaining Compute services.
The Compute cloud works using a series of daemon processes named
that exist persistently on the host machine. These binaries can all run on the
same machine or be spread out on multiple boxes in a large deployment. The
responsibilities of services and drivers are:
A server daemon that serves the Nova Metadata API.
A server daemon that serves the Nova OpenStack Compute API.
A server daemon that serves the metadata and compute APIs in separate greenthreads.
Manages virtual machines. Loads a Service object, and exposes the public methods on ComputeManager through a Remote Procedure Call (RPC).
Provides database-access support for compute nodes (thereby reducing security risks).
Dispatches requests for new virtual machines to the correct node.
Provides a VNC proxy for browsers, allowing VNC consoles to access virtual machines.
Provides a SPICE proxy for browsers, allowing SPICE consoles to access virtual machines.
Provides a serial console proxy, allowing users to access a virtual machine’s serial console.
The architecture is covered in much greater detail in Nova System Architecture.
Some services have drivers that change how the service implements its core
functionality. For example, the
nova-compute service supports drivers
that let you choose which hypervisor type it can use.
There is information you might want to consider before doing your deployment, especially if it is going to be a larger deployment. For smaller deployments the defaults from the install guide will be sufficient.
Compute Driver Features Supported: While the majority of nova deployments use libvirt/kvm, you can use nova with other compute drivers. Nova attempts to provide a unified feature set across these, however, not all features are implemented on all backends, and not all features are equally well tested.
Feature Support by Use Case: A view of what features each driver supports based on what’s important to some large use cases (General Purpose Cloud, NFV Cloud, HPC Cloud).
Feature Support full list: A detailed dive through features in each compute driver backend.
Cells v2 configuration: For large deployments, cells v2 cells allow sharding of your compute environment. Upfront planning is key to a successful cells v2 layout.
Availability Zones: Availability Zones are an end-user visible logical abstraction for partitioning a cloud without knowing the physical infrastructure.
Placement service: Overview of the placement service, including how it fits in with the rest of nova.
Running nova-api on wsgi: Considerations for using a real WSGI container instead of the baked-in eventlet web server.
- Cells (v2)
- Host aggregates
- Compute service node firewall requirements
- Availability Zones
Once you have an OpenStack deployment up and running, you will want to manage it. The below guides cover everything from creating initial flavor and image to log management and live migration of instances.
Quotas: Managing project quotas in nova.
Scheduling: How the scheduler is configured, and how that will impact where compute instances land in your environment. If you are seeing unexpected distribution of compute instances in your hosts, you’ll want to dive into this configuration.
Exposing custom metadata to compute instances: How and when you might want to extend the basic metadata exposed to compute instances (either via metadata server or config drive) for your specific purposes.
- Manage the cloud
- Manage Compute services
- Configure Compute service groups
- Secure with rootwrap
- Configure SSH between compute nodes
- Configure live migrations
- Live-migrate instances
- Secure live migration with QEMU-native TLS
- Manage volumes
- Manage Flavors
- Injecting the administrator password
- Configure remote console access
- Compute schedulers
- Config drives
- Image Caching
- Metadata service
- Manage quotas
- Networking with neutron
- Manage project security
- Security hardening
OpenStack clouds run on platforms that differ greatly in the capabilities that they provide. By default, the Compute service seeks to abstract the underlying hardware that it runs on, rather than exposing specifics about the underlying host platforms. This abstraction manifests itself in many ways. For example, rather than exposing the types and topologies of CPUs running on hosts, the service exposes a number of generic CPUs (virtual CPUs, or vCPUs) and allows for overcommitting of these. In a similar manner, rather than exposing the individual types of network devices available on hosts, generic software-powered network ports are provided. These features are designed to allow high resource utilization and allows the service to provide a generic cost-effective and highly scalable cloud upon which to build applications.
This abstraction is beneficial for most workloads. However, there are some workloads where determinism and per-instance performance are important, if not vital. In these cases, instances can be expected to deliver near-native performance. The Compute service provides features to improve individual instance for these kind of workloads.
In deployments older than Train, or in mixed Stein/Train deployments with a
rolling upgrade in progress, unless
enabled, live migration is not
possible for instances with a NUMA topology when using the libvirt
driver. A NUMA topology may be specified explicitly or can be added
implicitly due to the use of CPU pinning or huge pages. Refer to bug
#1289064 for more information. As of Train, live migration of instances
with a NUMA topology when using the libvirt driver is fully supported.
- Attaching physical PCI devices to guests
- CPU topologies
- SMP, NUMA, and SMT
- PCPU and VCPU
- Customizing instance NUMA placement policies
- Customizing instance CPU pinning policies
- Customizing instance CPU topologies
- Configuring libvirt compute nodes for CPU pinning
- Configuring CPU power management for dedicated cores
- Configuring Hyper-V compute nodes for instance NUMA policies
- Real Time
- Huge pages
- Attaching virtual GPU devices to guests
- File-backed memory
- Using ports with resource request
- Using ports vnic_type=’vdpa’
- Attaching virtual persistent memory to guests
- Emulated Trusted Platform Module (vTPM)
- Secure Boot
- AMD SEV (Secure Encrypted Virtualization)
- Managing Resource Providers Using Config Files
- Compute Node Identification
- Resource Limits
- CPU models
- Other libvirt features
Once you are running nova, the following information is extremely useful.
Upgrades: How nova is designed to be upgraded for minimal service impact, and the order you should do them in.
- Troubleshoot Compute
- Orphaned resource allocations
- Rebuild placement DB
- Affinity policy violated with parallel requests
- Compute service logging
- Guru Meditation reports
- Common errors and fixes for Compute
- Credential errors, 401, and 403 forbidden errors
- Live migration permission issues
- Instance errors
- Empty log output for Linux instances
- Reset the state of an instance
- Injection problems
- Cannot find suitable emulator for x86_64
- Failed to attach volume after detaching
- Failed to attach volume, systool is not installed
- Failed to connect volume in FC SAN
- Multipath call failed exit
- Failed to Attach Volume, Missing sg_scan
- Requested microversions are ignored
- User token times out during long-running operations
- Evacuate instances
- Migrate instances
- Use snapshots to migrate instances
- Recover from a failed compute node
- hw_machine_type - Configuring and updating QEMU instance machine types
- hw_emulation_architecture - Configuring QEMU instance emulation architecture
- Soft Delete and Shadow Tables